Color Image Reproduction of Scenes With Color Enhancement and Preferential Tone Mapping

This invention relates to the art of producing a visual color reproduction of a scene. More specifically, the invention relates to a visual color reproduction of a scene having improved color and pleasing skin tones.

Color image reproduction methods and systems known in the art capture images on image-receptive media, which can be stored in analog or digital form, and then output as a visual reproduction. For example, color images may be captured on negative film and then reproduced optically or electronically on photographic paper. Images can also be captured on positive photographic materials, known as transparencies, and then viewed directly by projection or back-illumination, or copied onto other transparencies, or printed onto photographic paper. In other cases, color images can be captured by electronic devices, such as video or still CCD cameras, and viewed on monitors or printed using dye sublimation thermal printers. In each case previously cited, these systems are subjected to customer satisfaction criteria and may or may not embody digital tone reproduction manipulation or some form of color enhancement. The systems mentioned above are just some examples of color image reproduction systems. The application of this invention is not limited to the above systems, but can be applied to any system for producing color reproductions, such as the reproduction of reflection original images encoded and stored on a KODAK PHOTO CD.TM. imaging system.

One of the important criteria for viewer satisfaction in photographic reproductions is the correspondence between the optical reflectances of objects in the original scene compared to those of the reproduction. Viewers prefer to have high quality images with pleasing tone reproduction and high colorfulness while maintaining good skin tone. Technological advances have been made over the years in photographic films by incorporating more chemical enhancement, and in photographic papers by increasing the paper contrast. Some current methods for making color reproductions produce fairly bright colors and offer reasonable skin tone reproduction; however, there have been limitations on the extent to which color enhancement can be employed. One common problem associated with higher color enhancement films is that the reproduced colors are more saturated while the skin tone reproduction is oversaturated. The resulting reproductions are judged to be inferior to less colorful prints. Moreover, it has not been fully appreciated that the preferred visual reproduction does not usually correspond to the most accurate rendition.

Conventional silver halide photographic systems are subject to limitations imposed by optically printing one chemically developed material onto another chemically developable material and do not reproduce the scenes in a way that is preferred to the viewer. More specifically, conventional systems have not produced images having preferred highly saturated colors without adversely affecting desired skin tones.

Aside from color enhancement, the quality of image reproductions is also affected by the tone scale or tone mapping employed to reproduce the density variations that make up an image. It has previously been discovered that the use of a preferential tone scale or mapping as described generally in U.S. Pat. No. 5,300,381 issued to J. Buhr and H. Franchino entitled "Color Image Reproduction of Scenes with Preferential Tone Mapping", which is incorporated herein by reference in its entirety, can be utilized to provide a reproduced image that is perceived by the viewer to be a reproduction of the original scene which is preferred to that previously obtainable. The prior improvement in tone mapping has provided a degree of preferred reproduction of color images but the use of tone mapping alone has not enabled the full extent of improvement desired by the viewer.

It is a problem to be solved to provide viewer preferred visual reproductions of scenes which exhibit both increased color saturation and pleasing skin tones.

A system and method of image reproduction in color with preferential tone mapping and color enhancement are provided in which the method provides a reproduction of an original scene with enhanced color and with a preferential tone mapping, the method comprising:

(I) capturing the parameters of the original scene;

(II) transforming the original scene parameters, in conjunction with the untransfomed characteristics of the capturing and creating steps, wherein said transforming includes modifying the parameters to create a reproduction which satisfies the following two requirements:

(a) the reproduction has a reproduced tone mapping having instantaneous gamma values of viewed reproduction density relative to density of the original scene which are greater than a minimum value of A and less than A plus 0.35 times scene exposure density over a scene exposure density range of from 0.60 to 1.45 as measured relative to a zero density, 100% diffuse reflector in the original scene, the instantaneous gamma values within scene exposure density range further lying entirely within a range of values subtended by an angle of 14.degree. measured from a reference point at 0.0 scene exposure density and a viewed reproduction density gradient of A, wherein A is 1.0 in the case of direct view reflection media reproduction, 1.1 in the case of self-illuminated reproduction for viewing with ambient surround, and 1.3 in the case of a reproduction for viewing with dark surround; and

(b) the reproduction has color enhancement wherein a Macbeth Color Checker, if captured in the original scene, would appear in the reproduced image so that (1) the CIELAB chroma ratio is greater than 1/A for at least one of the six high chroma patches, and (2) either the CIELAB chroma ratio for the light skin patch is between 0.8/A and 1.4/A or the CIELAB chroma ratio for the dark skin patch is between 1.0/A and 1.8/A, or the CIELAB chroma ratios for both skin patches are within said ranges, wherein A is as defined in (a) above; and

(III) creating a reproduction of the original scene.

The CIELAB chroma ratios, which are defined more fully below, are the ratios for a given color patch of the CIELAB chroma (c*) of the reproduction prepared in accordance with the prescribed steps to the c* of the original scene.

The invention also encompasses a system for making the preferred visual reproduction and the reproduction so made.

The invention provides viewer preferred visual reproductions of scenes which exhibit both increased color saturation and pleasing skin tones.

In the drawings:

FIG. 1 is a graph illustrating image reproduction gamma and instantaneous gamma characteristics of a prior art photographic system using conventional amateur negative film and print paper.

FIG. 2 is a graph similar to FIG. 1 but for a prior art conventional professional negative film and print paper.

FIG. 3 is a graph similar to FIG. 1 illustrating a comparison of a variety of conventional image reproduction gamma and instantaneous gamma characteristics with the corresponding characteristics for a system and method of the present invention.

FIG. 4 is a graph illustrating scene image reproduction gamma and instantaneous gamma characteristics for a variety of image reproduction systems of the invention.

FIGS. 5-7 are graphs of boundary conditions for scene image reproduction instantaneous gamma characteristics that are useful in explaining and defining the parameters of the present invention.

FIG. 8 is a schematic representation of a scene image reproduction system in accordance with the system and method of the present invention.

FIG. 9 is a four quadrant tone reproduction diagram for the system of FIG. 7.

FIG. 10 is a schematic representation of an alternative scene image reproduction system in accordance with present invention.

FIG. 11 is a four quadrant tone reproduction diagram for the system of FIG. 9.

FIG. 12 is a schematic representation of an scene image reproduction system incorporating digital image storage on a compact disc and illustrating alternative configurations in which the present invention might be practiced.

FIG. 13 is a schematic illustration of a test arrangement for mapping tone reproduction of a scene image reproduction system.

FIG. 14 is a schematic for performing color enhancement.

FIG. 15 is a schematic for converting signals.

FIG. 16 is second schematic for converting signals.

FIG. 17 is a third schematic for converting signals.

FIG. 18 is a fourth schematic for converting signals.

It has been found that reproductions that have been modified as prescribed in the invention, with enhancement of the color saturation while maintaining pleasing skin tones, provide images that viewers prefer. The invention objective may be accomplished by employing electronic, chemical, or a combination of electronic and chemical methods to transform the original scene parameters into a modified set of parameters in the visual reproduction in which the modified parameters satisfy both a prescribed tone scale relationship to the original and a prescribed color enhancement relationship to the original.

Digital intermediary are particularly suited to allow tone and color enhancements to be made to the image. An improved digital color imaging system would have the capability to produce images having highly saturated colors and pleasing skin tone reproduction that viewers would perceive as a preferred reproduction of the original scene. These enhancements could also correct for variations in overall exposure and color balance due to exposure control errors on the image capturing device, variations in color temperature of the taking illuminant, and variations in the color responses of the output devices. The data representing the digital image can represent various metrics used to describe the image color and tone scale or can be related to the original scene. This enhancement could be performed in color spaces which include RGB, CMY, CIELAB, or scene exposure space, with corrections in the form of scalars, matrices, polynomials, or one- or multi-dimensional lookup tables while acknowledging the gamut of the output device in order to maintain the desired image response.

An improved digital imaging system could readily incorporate these enhancements; thus, images captured on negative or positive film could be shown on video monitors or printed on negative or positive photographic papers. These images could be printed using thermal, ink jet, electrophotographic, or any other means known in the art. These digitized images could also be stored for later display. Aside from standard calibration procedures, with regard to the stored images and their original capture medium, no adjustments to the devices need be required in order to make a preferred reproduction of the original image.

An unexpected advantage is gained by producing the visual reproduction to meet both the preferred tone scale mapping and color enhancement limitations of the invention. Thus a conventional or digital color imaging system could produce images that are not only aesthetically pleasing, but also appropriately rendered productions of any image using any media or device regardless of the origin or the media selected for the reproduction of the image.

The methods described permit reproduction of conventional and digitized images having tone and color reproduction characteristics that are preferred over those currently known and are perceived to be a preferred reproduction of the original scene. The scene may be reproduced as a reflection print by any means including a print process, such as a photographic, electrophotographic, inkjet or thermal dye transfer or the like process, or as a positive transparency or as a self-illuminated image, such as in the case of a video image.

The invention may be described in terms of its (I) tone mapping and (II) color enhancement aspects.

(I) Tone Mapping

The relationship between the visually perceived densities of objects in an image scene reproduction compared to those in the original scene is a critical aspect of achieving viewer satisfaction with the reproduction. Visually perceived densities of objects in an original scene or in an image reproduction of the scene include effects of viewing flare and, as is known in the art, can typically be measured using a low-flare radiometer. The theoretically correct reproduction for this purpose is generally considered to be a one-to-one relationship between the densities of the original scene and the densities of the reproduction. This is described in the book "The Reproduction of Colour" by Dr. R. W. G. Hunt, (Fountain Press, England--Fourth Ed.), specifically in Chapter 6 wherein the fundamentals of tone reproduction are discussed.

In reality however, actual tone reproduction produced by practical image reproduction systems deviates significantly from the theoretical one-to-one relationship that one might logically expect. For example, in conventional silver-halide based photographic systems, the silver-halide materials inherently tend to produce a well known non-linear, "S"-shaped relationship between the viewed print density (in the case of a print system) vs. scene exposure, such as shown on page 54 of the aforementioned Hunt reference. Similar curves are illustrated in FIGS. 1 and 2 in the accompanying drawings wherein FIG. 1 shows the D-logE curve D and its corresponding instantaneous gamma curve G representative of a typical amateur film/paper system and FIG. 2 shows the same class of curves D' and G' representative of a typical professional film/paper system. The term "instantaneous gamma" as used herein refers to the instantaneous rate of change of the image density, as perceived by the viewer, versus scene exposure density. Thus the curves G and G' correspond to the incremental slope values of the curves D and D', respectively.

It can be readily seen from FIGS. 1 and 2 that the instantaneous gamma curves of the reproduced image in both systems exhibit somewhat bell-shaped curvature in the mid-tone range with significant difference between the maximum and minimum values of the curves within the mid-tone range. The major result of this behavior in these reproduction systems is that the tone mapping of the reproduced image contains somewhat of an unnatural amplification of the mid-tone information in the image with a compression of the shadow and highlight information. The visual effect is one of a degree of harshness in the image that is recognized by the viewer as not being a natural reproduction of the original scene. It should be noted here that image reproduction of a scene that is perceived by a viewer entails consideration of the effect that any stage in the reproduction process has on the entire process from image capture to final reproduction of the image. In U.S. Pat. No. 4,792,518--Kuwashima et al, there is disclosed a silver halide color reversal reflection print material, the purpose of which is to faithfully reproduce an original image which might exist in the form of either a transparency image or a reflection print image. This disclosure describes a method for faithfully, i.e. exactly, copying a transmission or reflection original and stresses the need for linearity in the gamma curve of the reproduction material across an extended range of original image density to achieve a faithful reproduction of the original image source (slide or print). To this extent, it is similar to the basic concept of a one-to-one relationship mentioned above in connection with the Hunt publication. Such a concept however, does not recognize or suggest what is needed to provide a preferred color image reproduction of an original scene nor does it take into account characteristics of the image capture medium and viewing flare when viewing the reproduced image. The print materials disclosed in this patent would simply reproduce whatever gamma characteristics exist in the image origination materials, which at present are distinctly non-linear, without regard to whether such materials would result in a desired reproduction of the original scene.

The challenge in image scene reproduction, therefore, is to reproduce the original scene in such a manner that upon viewing the reproduction, the viewer is pleased with the reproduced scene even though it may not be exactly faithful to the original scene, i.e. does not present a measurable one-to-one density mapping relationship between the original scene and the reproduction. The result would then be a pleasing reproduction that would be preferred over reproductions that do not convey that impression.

The preferred tone mapping matches neither the theoretical 1:1 relationship nor the actual tone mapping relationship that is achieved with conventional image reproduction systems. Data developed in these tests have led to the conclusion that the best images are produced for most of a variety of scene types when the perceived densities of the reproduced image are related to the viewed densities of the original scene by an approximately linear function of instantaneous gamma over a large part of the scene density range. A comparison of representative conventional tone mappings with one example of a preferred tone mapping according to the invention is illustrated in FIG. 3 which shows a series of D-logE curves 10d-13d and corresponding instantaneous gamma curves 10g-13g for a variety of different film/paper hardcopy image reproduction systems. Curve pairs 10d, 10g-12d, 12g were generated from conventionally available negative film/print paper photographic systems while curve pair 13d, 13g, which illustrates the preferred tone mapping, was generated using conventional negative film optoelectronically scanned to develop pixel signals which were then image-processed and printed on color print paper. The general techniques of optoelectronic image scanning, digital signal processing and digital color printing are well known and do not, of themselves require detailed discussion herein. However, the particular image transformation function employed in the digital image processing was adapted in accordance with a feature of the invention to introduce a transformation that resulted in the perceived tone mapping characteristic shown by curves 13d,13g in FIG. 3. It will be appreciated that each of the curve pairs 10d, 10g through 13d, 13g represent different system-dependent reproduction characteristics in which the resultant output tone mapping relative to scene input information is dependent on the functional characteristics of the materials employed in the scene reproduction process. It will also be noted that for the three systems represented by curve pairs 10d, 10g through 12d, 12g, the instantaneous gamma curves 10g-12g are substantially bell-shaped in the mid-tone rage with significant variation between the maximum and minimum values of the curves within the mid-tone range. The mid-tone range, for this purpose can be seen to extend from about 0.6 to about 1.45 relative to a zero density, 100% diffuse reflector in the original scene. This reference reflector is represented in the graph of FIG. 3 by point 18 at the intersection of zero exposure density and 1.0 image reproduction density gradient and is commonly referred to in the art as a reference of 100% white in the scene. By comparison with the conventional systems, however, the instantaneous gamma curve 13g for the system of the invention is more flat, without the bell-shaped characteristic, approaching that having a nearly horizontal, linear segment in this same mid-tone range of scene exposure densities of from about 0.6 to about 1.45 relative to 100% white in the scene. While a single system according to the invention, as represented by curves 13d, 13g, is illustrated in FIG. 3, it has been found through the aforesaid empirical tests that perceived natural image reproduction is achieved by systems that exhibit values of instantaneous gamma falling within a range of output densities, albeit fairly limited in extent, as is graphically indicated in FIG. 3 by trapezoidal outline or window 15. This graphically illustrated range of instantaneous gamma values can be described in quantitative terms as being a range of values of viewed reproduction density gradient relative to density of the original scene which are greater than 1.0 and less than 1.0 plus 0.35 times scene exposure density over a scene exposure density range of from about 0.60 to about 1.45 as measured relative to a zero density, 100% diffuse reflector in the original scene. In FIG. 4, a number of instantaneous gamma functions according to the invention are shown which fall within this described range and which have been found to give preferred perceived image reproduction for a wide variety of scene types.

It has been found that, for most scenes, preferred image reproduction results in gamma functions that fall within the lower portion of the range 15. This range of preferred gamma function is illustrated by trapezoid 15a in FIG. 5 and may be described quantitatively as being viewed reproduction density gradient values greater than 1.0 and less than 1.0 plus 0.25 times scene exposure density in the range of scene exposure density of from 0.6 to 1.45 measured against the 100% white as described above. For some scenes, however, it has been discovered that a slightly modified range of instantaneous gamma values give preferred images. Scenes of the latter type are generally scenes which have limited contrast range such as low light scenes; a dusk scene or a foggy scene being typical examples. For scenes of this type, preferred instantaneous gammas would fall in the upper portion of the range 15 in FIG. 4 and as illustrated specifically in FIG. 6. While these latter instantaneous gammas are slightly higher overall than for the "most scenes" range of instantaneous gammas shown in FIG. 5, the included range of values in each boundary set is approximately the same. Thus further in accordance with the invention, the range of acceptable gamma values within the aforementioned mid-tone range may be further defined as lying entirely within a range of output density values subtended by an angle of about 14.degree. measured from the 100% white reference point 18. This is illustrated in FIG. 7 in which the solid line trapezoid 15a represents gamma range boundary conditions that provide preferred viewed image reproduction for "most scenes" case while the dotted line trapezoid 15b is applicable to low contrast scenes. In each case, as can be seen, the angle subtended by the trapezoidal area is indicated as being 14.degree.. While 14.degree. is indicated as the preferred boundary condition, it will be appreciated that some limited deviation from this precise figure is possible. Techniques and apparatus for film image classification in photofinishing processes are well known. Thus, it is within the scope of the invention to provide for image classification, for example in film scanner apparatus 24, to identify images which are best reproduced with the instantaneous gamma characteristic 15a and other scenes, notably low contrast scenes, which are best reproduced with the instantaneous gamma characteristic 15b. Other scene classifications can be envisioned that would preferably use an instantaneous gamma characteristic range as described above other than 15a or 15b but nonetheless falling within the overall boundaries of trapezoid 15. Knowing the image classification, the appropriate lookup table in image processor 26 would then be employed to achieve the desired resultant instantaneous gamma characteristic in the image reproduction.

In the foregoing and succeeding discussion regarding the Figures, it has been assumed that the reproduction to be viewed is one on reflective media for direct view and therefore A, as described in the Summary of the Invention, is 1.0. In the case of a reproduction that is self-illuminated for viewing with ambient surround or one that is for viewing with dark surround, the value for A would be 1.1 or 1.3, respectively.

Up to this point, the invention has been described with emphasis on the nature of the instantaneous gamma function of the viewed image reproduction. This is due in large part to the fact that the reproduction of an original scene involves a knowledge and understanding of the manner in which scene parameters are affected by each step in the process of scene capture and image reproduction. With this newly discovered knowledge of what constitutes preferred image reproduction, it then becomes possible to direct the adjustment of any one or more of the stages in the overall process to achieve the desired result of natural image reproduction which is best characterized in a quantitative manner by the instantaneous gamma function definitions set forth above. With this knowledge, many methods and corresponding systems can be formulated to achieve this result. Some examples of such systems will now be considered.

Referring to FIG. 8, an example image reproduction system is shown which incorporates provision for producing the preferred tone reproduction images in accordance with the present invention. In this system, a conventional single lens reflex camera 20 may be used to capture a scene 19 on conventional ISO 100 color negative film (not shown). This film is then processed in known manner in a film processor 22 to produce a negative film image having a D-logE characteristic curve 40 illustrated in Quadrant I of the four quadrant tone reproduction diagram of FIG. 9. It will be appreciated by those skilled in the art that the film D-logE curve 40 includes the effects of the camera-exposing flare introduced by virtue of light scatter in the camera and taking lens. Typically, the film from the processing stage would then be directly printed at the photofinishing lab using a printer and a color print paper processor 30 to produce a hardcopy color print 32. The transfer function of such conventional print production might typically be represented as shown in Quadrant III of FIG. 9. Desirably, the print 32 should appear to a viewer 34 as a preferred natural reproduction of scene 19. However, assuming the scene capture medium, as represented by the camera and film characteristics of Quadrant I, and the image reproduction medium, as represented by the printer and paper characteristics of Quadrant III, are all conventional, the resultant D-logE and corresponding instantaneous gamma function will typically appear as shown by one of the conventional curve pairs 10d, 10g-12d, 12g of FIG. 3. In other words, the instantaneous gamma function for the image reproduction would have substantial portions lying outside the preferred trapezoidal tone mapping range within the boundaries of trapezoid 15. However, even with the photographic system such as just described involving conventional film and paper processing, it is possible to shape the resultant D-LogE and corresponding gamma function curves of Quadrant IV by providing a predetermined digital or analog transformation function, as represented in Quadrant II, which, taken in conjunction with the conventional transfer functions of Quadrants I and III, will achieve the desired output function of Quadrant IV. Thus, to this end in the embodiment of FIG. 8, the developed film is applied to a film scanner 24 in which the image is converted to a series of image pixel signals, the amplitude value of each pixel signal being related to the density of the corresponding pixels in the film image. The image pixel signals are then applied to an image signal processor 26 wherein the image pixel signals are modified in accordance with the predetermined transformation characteristic of Quadrant II and fed to an appropriate printing device such as a three color laser printer 28 used to expose the photographic paper prior to being processed in processor 30. Preferably, the scanned pixel signals are converted to digital form in known manner, either in the scanner 24 or the signal processor 26, to simplify the signal modifying processing function in the processor 26. Since the transfer characteristics of the scene capture and image reproduction media are easily measured and generally well known, it is a fairly straightforward matter utilizing known techniques in the art of digital image processing to establish the intermediate transfer function of Quadrant II in the form of a lookup table in the digital processor 26 to achieve the desired results. The advantage of using digital image processing as just described is the ability to achieve the preferential tone mapping of the invention using image capture and image reproduction media which do not inherently have the desired transformation functions. Alternatively, it's possible to embody the desired transformation function within the material of either the image capture stage, e.g. in the photographic film or within the image reproduction stage, e.g. in an intermediate film stage or in a photographic print paper, in the case of photographic reproduction systems. It is also possible to share the transformation function within both the image capture and image reproduction stage as might be the case, for example, where matched film and paper media are employed.

FIG. 10 illustrates an alternative system involving a purely electronic image capture system including an electronic camera 50 and an RGB image signal store 52. In a system of this type, the image is typically captured directly in the form of image pixel signals which may then be manipulated in an image signal transformation processor 54, in accordance with the invention, using an appropriate lookup table to modify the image signals according to the transfer function of Quadrant II in FIG. 11. The modified signals are then applied to a conventional color printer 56 where the image is printed directly in hardcopy form in known manner using thermal dye transfer, inkier processes or laser printing or the like. The transfer function of Quadrant I corresponds to the conventional transfer function of the scene capture stage, including camera 50, and the transfer function of Quadrant III corresponds to the conventional transfer function of the image reproduction stage, including printer 56. It will be appreciated, however, that the image processor transfer function modification of Quadrant II can also be incorporated into either the scene capture stage or the image reproduction stage thus modifying the corresponding transfer function of Quadrant I or Quadrant III, respectively, such as was described in respect to the photographic system of FIG. 8.

Referring now to FIG. 12, there is shown an image reproduction system of the hybrid type involving capability for both photographic and electronic image reproduction. A system of this type, as represented by the solid line functional blocks, has been introduced by the Eastman Kodak Company under the name of PHOTO-CD and is capable of providing reproduced images in a variety of formats such as by means of photographic prints and video images on the screen of a television cathode ray tube. As illustrated in FIG. 12, a photographic camera 60 is used to expose film which is submitted to a photographic lab where it is developed in a film processor 62 and applied to a print processor 64 for the production of photographic prints 66. In accordance with the principals of the aforementioned PHOTO-CD process, the film negatives may also be applied to a film scanner 68 for generation of image pixel signals which are converted to digital form for recording on an intermediate signal recording medium such as an optical recording compact disc (not shown) by means of a CD recorder 70. The film negatives, prints and compact disc are then furnished to the customer. Assuming the customer has an appropriate CD player 72 connected to a television set 74, the images recorded on the disc can be played back through the TV set for viewing, and manipulating in a number of ways, such as zooming, cropping etc. Alternatively, the image signals output from the CD player 72 may be applied to a hardcopy printer 76, which may, for example, comprise a thermal dye transfer color printer, for the production of color prints 78 in the home with or without manipulation of the prints, i.e. zooming, cropping etc.

In accordance with the invention, the tone mapping in the output reproduced image may be accomplished by modifying the image pixel signals as described above using a signal processor 80 at the output of the film scanner 68 in the photographic lab prior to recording of the signals on the compact disc or by means of a signal processor 82 or 84 at the output of the disc player 72. In practice, the processors 82 and 84 might be the same with provision for selection of the appropriate lookup table for the desired transfer function depending on whether tone mapping adapted for the thermal printer or for the video image produced in the TV set 74 is to be employed. Adaption for images reproduced on a self-illuminated image reproduction device such as a TV cathode ray tube would be required due to the well known effect that the surround has on a viewer's perception of the image being viewed. This would be accommodated in the video image by increasing the reproduction contrast by 10% as illustrated by shifting the boundary conditions and the 100% scene white reference point of FIGS. 5-7 by +0.1 density gradient units to thereby produce parameters for a TV image having ambient surround equivalent to those for viewing a reflection print. In a similar vein, it will be appreciated that projected images, slide projections or video image projections, are generally viewed in relatively darkened surround conditions and equivalent parameters for projected images would be arrived at by shifting the boundary conditions and 100% white scene reference point of FIGS. 5-7 by +0.3 density gradient units.

As described in the invention summary, the tone mapping from original scene to preferred reproduction is comprised of these conceptual steps: (1) original scene parameter capture, (2) effecting desired transformation between the original scene parameters and the visual reproduction parameters, and (3) creation of the visual reproduction. The transformation may be effected through one or more steps between capture and reproduction or may be integrated with one or both of the capture and reproduction steps. The result of this process is a predetermined reproduced tone mapping in which instantaneous gamma values of viewed image reproduction density relative to density of the original scene are greater than A and less than A plus 0.35 times scene exposure density over a scene exposure density range of from 0.60 to 1.45 as measured relative to a zero density, 100% diffuse reflector in the original scene, the instantaneous gamma values within scene exposure density range further lying entirely within a range of values subtended by an angle of 14.degree. measured from a reference point at 0.0 scene exposure density and A viewed reproduction density gradient, wherein A is 1.0 in the case of direct view reflection media reproduction, 1.1 in the case of self-illuminated reproduction for viewing with ambient surround, and 1.3 in the case of a reproduction for viewing with dark surround.

Image manipulation consists of the altering of the tones in the captured scene in a way which will produce the desired tones in the final viewed image. Reproduction generation defines the creation of the final reproduction of the original scene which will be viewed. The process of capturing an image and generating a reproduction with the prescribed tonal mapping may be accomplished by a single complex element, as in instant photography, or in two or more separate steps. Regardless of the number of steps, conceptually the scene is first captured by a light sensitive element, transformed, and converted into a viewed material. For the "instant" process, the transformation between the tones in the original scene and their prescribed analogs in the reproduction is incorporated in the light sensitive element/display material. The tone mapping characteristics of each step in the process must be known so that all manipulations achieve the prescribed mapping between the original scene tones and those in the viewed reproduction.

In the following discussion, "scene" refers to the original scene parameters; "camera" refers to a device capable of controlling the optical exposure of a light sensitive element to capture a scene; "sensor" refers to a light sensitive element capable of capturing the tones in a scene in a quantifiable manner; "preferred tone mapping" refers to tone mapping in which instantaneous gamma values of viewed image reproduction density relative to density of the original scene is greater than A and less than A plus 0.35 times scene exposure density over a scene exposure density range of from 0.60 to 1.45 as measured relative to a zero density, 100% diffuse reflector in the original scene, the instantaneous gamma values within scene exposure density range further lying entirely within a range of values subtended by an angle of 14.degree. measured from a reference point at 0.0 scene exposure density and a viewed reproduction density gradient of A, wherein A is 1.0 in the case of direct view reflection media reproduction, 1.1 in the case of self-illuminated ambient surround reproduction, and 1.3 in the case of a dark surround reproduction; "storage" refers to the storage of scene parameters in any way which allows retrieval without loss of fidelity and includes, for example, electronic, optical, magnetic, magneto-optical, biological, chemical, and other mechanisms which provide reliable scene parameter storage.

The present invention may be employed in combination with features described in published materials relating to photographic processes including the following, which are incorporated herein by reference in their entirety:

1. R. W. G. Hunt, The Reproduction of Color in Photography, Printing, and Television, 4th Ed., 1987, Fountain Press, England.

2. Larish, John J., Understanding Electronic Photography, ISBN 0-8306-8315-4

3. E. Giorgianni, U.S. Pat. No. 5,267,030

4. Larish, Ibid, Chap. 4.

5. Holt, A., Electronic Circuits, ISBN 0-471-02313-2

6. Digital Image Processing, R. Gonzalez and P. Wintz, Addison-Wesley Publ. Co., 1977.

7. Digital Image Processing, W. K. Pratt, J. Wiley & Sons, 1991.

8. Fundamentals of Digital Image Processing, A. K. Jain, Prentice Hall, 1986.

9. Linear Systems, Fourier Transforms, and Optics, J. D. Gaskill, J. Wiley & Sons, 1978.

10. The Theory of the Photographic Process, T. H. James, Editor, 4th Edition, 1977, Macmillan Publishing Co., Inc.

11. "Device-Independent Color Imaging and Imaging Systems Integration", Proceedings of the SPIE, Vol. 1909, Feb., 1993.

12. "Color gamut mapping techniques for color hardcopy images", T. Hoshino and R. S. Berns, Proceedings of the SPIE, Vol. 1909, p 152, Feb., 1993.

13. Castellano, Handbook of Display Technologies, Academic Press, 1992-ISBN-0-12-163470-5.

16. J. Millman, J., Microelectronics, ISBN 0-07-042327-X

17. KLI2103 Performance Specification, Eastman Kodak Co.

18. (PIW) Hubbard, G., "PHOTO CD Systems: A Technical Overview", Proceedings of IS&T 44th Annual Conference, Advanced Printing of the Paper Summaries for the Proceedings of IS&T 44th Annual Conference, ISBN 0-89208-156-2, IS&T, May 12-17, 1991.

19. (Johnson, PCD storage format) Melnychuk and Barry, "PHOTO CD Systems: A Technical Overview", Proceedings of IS&T 44th Annual Conference, Advanced Printing of the Paper Summaries for the Proceedings of IS&T 44th Annual Conference, ISBN 0-89208-156-2, IS&T, May 12-17, 1991.

20. S. Thurm, K. Bunge, and G. Findeis, "Method of and Apparatus for Determining the Copying Light amounts for Copying from Color Originals, U.S. Pat. No. 54,279,502, Jul. 21, 1981.

21. Wheeler, Wilson, and O'Such, U.S. Pat. No. 5,049,916, Automatic "Optimization of Photographic Exposure Parameters through Determination and Utilization of Extra System Speed".1

22. Baumeister, U.S. Pat. No. 4,739,409, "Intelligent Exposure Control for Electronic Cameras".

23. C. J. Bartleson and R. W. Huboi, "Exposure determination methods for color printing: The concept of Optimum correction level", J. SMPTE, Vol. 65, p 205-215, 1956.

24. "Advanced Color Printing Technology for Photofinishers and Professional Finishers", Eastman Kodak CO., 1979.

25. R. M. Evans, "Method for Correcting Photographic Color Prints", U.S. Pat. No. 2,571,697, Oct. 16, 1951.

26. H.-C. Lee, U.S. Pat. No. 4,663,663, 1987, "Digital Color Image Processing Method Employing Constrained Correction of Color Reproduction Function".

28. Tannas, Flat Panel Display and CRT's, ISBN 0-442-28050-8

29. Retrieval of sensor recorded information by non-standard techniques such as electrical scanning of silver halide latent image, L. Kellogg, U.S. Pat. No. 4,788,131.

30. Research Disclosure, November 1992, Item 34390 published by Kenneth Mason Publications, Ltd., Dudley Annex, 12a North Street, Emsworth, Hampshire P010 7DQ, ENGLAND.

31. Research Disclosure, December 1989, Item 308119 published by Kenneth Mason Publications, Ltd., Dudley Annex, 12a North Street, Emsworth, Hampshire P010 7DQ, ENGLAND.

Step (1)--Original Scene Parameter Capture

Capturing the original scene parameters can be accomplished by any light sensitive element, or sensor, capable of sensing the tonal values of the objects in a scene in a manner which quantitatively determines their relative log luminances. The sensor is typically contained in a device or camera which controls its exposure to light. Examples of cameras and sensors include but are not limited to cameras using photographic films and electronic cameras using CCD (Charge-Coupled Device) sensors. Cameras and sensors may be of any suitable physical dimensions.

The sensors in conventional photographic applications are silver halide based photographic materials that may be negative or positive working film, semi-reflective film, or reflection paper although the most common embodiment is color negative film. Conventional films and papers contain a multilayer structure with separate color record imaging materials. Films specifically designed for scanning are also appropriate. The film may or may not be contained in spools, cartridges, or similar containers, depending on the type of camera in which it is used. The film may contain non-optically sensitive materials, such as magnetic or electrical elements. Scene capture may be accomplished using all currently available silver halide photo-sensitive films and papers as well as those that will appear in the future. In the case of radiation sensitized goods which capture via latent image, the capturing step encompasses the formation of the latent image and the image obtained upon development of the latent image.

All cameras designed for the use of silver halide photo-sensitive materials are suitable for capturing original scene parameters. Preferred cameras are those which deliver as uniform an illumination field as possible on the photo-sensitive element, producing a uniform capture of the scene. The camera may record two-dimensional or three-dimensional images which may be still or moving. A typical camera is of the 35 mm single lens reflex type.

In silver halide photography, the scene is captured and stored as latent image in the silver halide emulsion grains. Subsequent to or concerted with photosensitive element exposure, the recorded information is often converted to a more permanent representation of the original scene parameters by some process, typically chemical although it can be also be thermal, magnetic, optical, or electrical in nature. The resultant recorded image is typically a transparent film, although it may be a reflection material or any other storage medium. A typical embodiment for conventional silver halide photography is the use of a 35 mm single lens reflex camera exposing color negative film followed by Kodak.TM. Flexicolor.TM. C-41 chemical development to produce an optical density varying representation of the original scene.

It is possible to convert scene parameters captured by the exposed element directly to an analog or digital electronic representation by scanning the exposed photosensitive element with a scanner incorporating an electrical, chemical, or thermal process capable of reading out the latent image distribution recorded in the film or paper. This mechanism for scene parameter capture is also suitable for use to create the preferred reproduction.

The scene may also be captured with an electronic camera containing one or more photosensitive elements which may be solid state sensors, photomultiplier tubes, storage phosphors, or other materials performing the same function. Electronic cameras capture original scene parameters using a regular array of picture elements (pixels) in one or more sensors. Electronic photography useful in the invention is described in detail in Ref. 2. Solid state sensors include CCD's, typically photocapacitors (full frame sensor) or photodiodes (interline sensors). Interline CCD's can perform electronic shuttering.

Electronic cameras consist of a lens element and shutter which focus the scene onto the sensor and control the exposure level and duration of the sensor. The camera may capture two-dimensional or three-dimensional images in an analog or digital mode which may be still or moving. A camera of this type will sense the tones in the original scene and convert them into an appropriate form for signal storage. The signal may be stored magnetically, optically, electronically, chemically, or biologically in the camera itself or associated equipment. A suitable embodiment for electronic capture is a Kodak DCS electronic camera which captures a scene with a solid state full frame photocapacitor CCD sensor and buffers the image in electronic RAM and stores the image on a magnetic hard disk located in the camera. The image data in RAM may be transmitted and then stored on media, such as magnetic tape, optical disk, magnetic disk, magneto-optical disks, etc.

Storage of scene parameters may be accomplished in a variety of ways. Electronic input may consist of image data representation stored using magnetic, optical, magneto-optical, RAM, biological, solid state, or other materials which permanently or semi-permanently record information in a retrievable manner. The mode input storage may be analog or digital. Magnetic storage is exemplified by floppy disks, hard drives, video tape, etc. on which are stored the representation of the original scene parameters stored as oriented magnetic domains which are written and retrieved by using magnetic heads. One suitable embodiment of magnetic storage is the PLI Infinity 88 RW/44 removable hard drive. Examples of optical storage include optical disks, laser disks, etc. on which the representation of the original scene parameters are stored in binary form as ablated pits, which when read reflectively result in variations in optical density at some wavelength on the optical media. The scene representation is recorded and retrieved by optical radiation means such as lasers. A suitable embodiment of optical storage is the PHOTO CD disk. Magneto-optical storage devices use different technologies for storage, including both optical and magnetic mechanisms, for both reading and writing. A suitable example of magnetic recording on rerecordable media is the PLI Infinity 21 MB Floptical Drive that uses magnetic writing and reading along with an optical servo for precise alignment. A suitable example of optical reading and writing in a magneto-optical device is the DataPak MO/650 optical drive that uses an applied magnetic field during the optical writing process. Examples of solid state storage include EPROMs (Electrically Programable Read-Only Memory), EEPROMs (Eraseable EPROM), and PROMs (Programmable Read-Only Memory), Ferroelectric nonvolatile RAMs (Random Access Memory), magnetic bubble memory, California Institute of Technology's biological chip NMOS (n-channel metal-oxide semiconductor) memory "ASSOCMEM", etc. This type of storage uses solid state devices to store image pixel information by electrical or magnetic means. The recording and reading of the stored information is effected by electrical signals. A suitable embodiment of solid state storage is a PCMCIA (Personal Computer Memory Card International Association) flash EEPROM and other storage techniques described in Ref. 2, Chapter 4.

Storage of scene parameters in analog or digital form requires using a storage format. This format specifies how the image is stored and it is device dependent. For example, an image may be stored in accordance with the JPEG (Joint Photographic Equipment Group) compression standard in flash EEPROM or in a PHOTO CD format in a Kodak PHOTO CD.TM. disk (optical storage.) This format must be known by the writing and reading operations to accurately retrieve the stored scene parameters. A suitable embodiment for digital image formatting is the optical PHOTO CD.TM. disk of Eastman Kodak Co. (Ref. 9)

Instead of direct capture of the original scene parameters, it is also possible to access a representation of the original scene parameters, captured and stored at some prior time. These representations may be two-dimensional or three-dimensional and may be of still or moving scenes. The only requirement for this means of generating a preferred viewed reproduction of the original scene is that the relationship between the original scene parameters and those in the accessed original scene representation be known or that it be possible to make an accurate assumption about this relationship. The accessed scene representation was at some point captured using the methods described above for direct original scene parameter capture. A suitable embodiment for producing original scene representations is the PHOTO CDCD system taught by Giorgianni (Ref. 3).

In addition to the typical pictorial original scene capture, the original scene may be an original creation (computer generated or otherwise) to be accessed through information storage or it may be a computer simulation of an actual scene. It may also be comprised in whole or part of text material or graphic designs or patterns. Any of the types of stored scene parameters as described above are acceptable as input as long as the relationship between the original scene and stored scene representations are known.

Computer simulations are also acceptable input representations of original scene parameters as long as the relationship between the original scene parameters and those in the accessed original scene representation are known or that it is possible to make an accurate assumption about this relationship. Computer simulations are based on an original scene capture as defined above. The computer simulation may manipulate any aspect of the original scene parameter, such as color, sharpness, contrast, etc. as long as the visual mapping modifications to the original scene parameters are quantified. The computer simulation scene representation is usually stored in computer RAM but may be stored on any of the storage media defined above. Examples of computer simulations include scene manipulation by Adobe Photoshop, Kodak Premier scene manipulation, or other similar readily available software which manipulates the scene parameters in a quantifiable manner.

Converting optical representations of original scene parameters to electronic representations

Electronic representations of original scene parameters can be created by converting non-electronic representations. For example, exposed and processed color negative film can be optically read using a microdensitometer to produce a digitized image. Any optical representation can be transformed in some manner to produce an acceptable intermediate representation of the original scene parameters as long as the transformation step is conducted in a quantifiable manner. Appropriate photographic media include transparent films, semi-reflective films, and reflective papers, both positive and negative working. These optical representations may be two-dimensional or three-dimensional and either still or moving scenes.

Optical scanning is an example of the conversion operation. Scanning can be accomplished by using a microdensitometer, linear CCD array, or similar device. The scanning of two-dimensional and three-dimensional optical representations can be done in transmission or reflection mode with a device which scans by point, line, or area. A microdensitometer generally uses a small illuminated spot to measure with 3 colored filters the point-by-point (pixel-by-pixel) transmission of a film sample or the reflection of a paper sample. The scanning process generally follows a regular pattern to completely measure the optical representation of the scene. The electronic signal measured by the microdensitometer or CCD scanner is usually an analog signal. If desirable, the initial analog signal may be digitized as an integral operation of the scanning device or may be effected as a subsequent step to the scanning operation. One suitable embodiment for film scanning is to measure exposed and processed color negative films with a Perkin-Elmer microdensitometer.

It is most convenient to use a CCD scanner to perform the transformation from optical representation to electronic representation because it can be done rapidly. A CCD scanner can scan a line of pixels at the same time or the entire image area with a 2-dimensional CCD sensor. CCD scanning techniques are described in Ref. 17 and Ref. 2. A suitable embodiment for CCD scanning is the Kodak KLI-2103 linear image sensor implemented in the Kodak PHOTO CD Imaging Workstation (Ref. 18).

The scanned scene representation may be stored using any of the storage media described above. The representation may be stored as analog or digitized signals.

Analog/Digital Conversions

Signal processing may involve analog or digital signals. Analog signals may be sampled to produce a digitized or digital signals by using an analog-to-digital conversion techniques such as sigma-delta modulation, parallel comparator, integrating converters, and successive approximation, dual slope, or binary counting with error feedback. An analog signal is transformed into a digital representation by assigning all values of the analog signal between two values to a single digital value. The almost infinite values available in the analog signal are mapped to a finite number of values in the digital signal. The reduction in the number of possible scene parameter values is frequently beneficial for digital image processing. However, the analog signal must be digitized without losing valuable scene information. Thus the digital quantization must be conducted with enough resolution to maintain the image quality. It is typical to employ 256 gray levels or more to achieve good image detail but a far lesser number can be employed.

Converting digital signals to analog signals is simpler using hardware and can be accomplished by error feedback or comparator techniques or by summing binary-weighted currents, by means of resistor ladders, etc. This kind of conversion involves selectively applying the analog voltage which is to be digitized to a set of resistors whose resistances are each twice as large as the one before it and comparing the output voltage to the input which attempts to recreate the best analog signal from the digital signal. An example of such a conversion is the generation of images using LCD (Liquid Crystal Display) technology. Ref. 16 discloses other digital/analog transformation methods useful in the invention.

Digital signals can also be converted to other digital signals. For example, an image representation consisting of 1024 gray levels may be sampled down to an image consisting of 256 gray levels. This kind of conversion can be useful in reducing image processing time.

It is also possible to convert one analog signal into another analog signal. An example is the conversion of analog RGB signals to analog NTSC (U.S. television standard) signals for display on a CRT (Cathode Ray Tube).

Step (2)--Effecting Desired Transformation Between Original Scene Parameters and the Visual Reproduction Parameters

Signal Processing

Signal processing is a key step in the transformation from the original scene parameters to the preferred visual reproduction parameters. Different kinds of signal processing are appropriate for the various processes which result in the obtaining the preferred viewed reproduction of the original scene. A suitable embodiment for creating digital reproductions from color negative film scene parameter capture is the use of the Kodak PHOTO CD system, which involves (1) scanning the exposed and processed color negative film, (2) transforming the digitized signal into standard PHOTO CD format using a Kodak PIW, (3) manipulating the PHOTO CD image with digital signal processing so that a Kodak XL7700 printer creates a viewed reproduction of the original scene which has the desired scene mapping (FIG. 12). But this is just one method for creating the preferred scene reproductions.

Signal processing may be applied to two-dimensional dimensional or three-dimensional image representations which are still or moving. Signal processing can be done in the optical, digital, and chemical domains (Ref. 5-10). It is possible to use any or all of these techniques to achieve the preferred viewed reproduction, depending on the pathway for original scene capture parameters to preferred reproduction creation. For example, a 6.times.7 cm format camera can be used to capture original scene parameters with two dimensional color negative film. During the chemical development of the film, the developer works with the silver halide emulsion and incorporated image modifying couplers in the film to create a chemically enhanced optical representation in the negative of the scene in both sharpness and color. The negative can then be digitized by a CCD linear array in a Kodak Premier scanner to create a digital image of the scene. The digital image can then be transformed using the Premier workstation so that the desired preferred view reproduction will be created when a Premier output 4.times.5 negative film is optically printed onto 16.times.20 color negative paper after first creating an unsharp silver mask from the 4.times.5 negative. This example is fairly common for the use of the Kodak Premier system except that the sharpening is often done digitally by the workstation.

Signal processing may be performed by the camera as by electronic cameras, during film development as in the case of chemical processing, associated with the film scanning, in computers after captured images are pixelated, and as part of the final image generation process. The processing can occur in any single or any combination of the above categories.

Chemical Processing

Chemical processing is an inherent part of the development of silver halide imaging materials. During the image development step, the developer can work with the chemical agents incorporated in the film or introduced in the processing solution to enhance the recorded image affecting the tone, color, sharpness, and graininess of the resultant recorded image. These effects can also be introduced by using intermediate, internegative, and interpositive film in any number or combination of sequential steps with associated chemical development. For example, the original scene parameters can be captured in latent image in Kodak Lumiere color positive film in a 35 mm camera. During E-6 chemical processing of the film, the developer works with agents incorporated in the film to produce a sharpened and color enhanced film. The positive film image is then contact printed onto a color internegative film to reverse the image and alter the tone mapping of the color positive film recorded image and to correct some color inaccuracies in that recorded image. The internegative film image is then contact printed onto an intermediate film which alters the tone mapping of the recorded image so that when the intermediate film image is printed in a color enlarger onto Kodak Edge color paper, the preferred view reproduction is created along with sharpening and color enhancement.

The chemical enhancement and tone manipulation can be accomplished in the so called instant photography case in a concerted process. In instant photography, the latent image scene recording occurs during the camera exposure step. The exposed film is immediately or subsequently processed by a developer or developers incorporated into the film packaging. The developing process works with the exposed latent image representation of the original scene to create and alter the recorded information to produce the visual reproduction having the desired tone scale directly in the film emulsion layers or by migration of imagewise released colorant. This process can occur with or without enhancement of color or sharpness in the final image.

Optical signal processing

For original scene parameters which are recorded in a manner which has the exact spatial relationship between scene elements as in the original scene, such as a photographic film exposure, it is possible to do optical processing to achieve the desired visual reproduction as well as to enhance other characteristics of the captured image such as sharpness. For example, when the original scene parameters are captured on color negative film, the final visual reproduction is usually created by optically printing the negative, with an imaging lens or by contact, onto color negative paper. It is possible to use intermediate photographic materials to alter the original scene parameters to produce a preferred viewed reproduction. These intermediates include masks, both sharp and unsharp, which are generally used to adjust the tone reproduction and reproduction sharpness, as well as intermediate, internegative, and interpositive films which are used to adjust different scene parameters such as color, tone reproduction, and sharpness as well as to introduce special effects using blue matte and montage. These intermediate, internegative, and interpositive steps can be combined in any number and any order to achieve the desired film viewed image.

In addition, there are spatially varying optical processing effects which can be introduced during the non-contact printing step such as dodging and burning where different parts of the scene receive different amounts of exposure due to preferentially blocking the printing light beam between the film negative and paper. Ref. 1 contains a thorough treatment of these effects. It is recognized that while in the example given here the processing is done optically, all of these transformations can be effected digitally as well. These techniques can be effected manually using software packages like Adobe Photoshop.

Optical transformations affecting reproduction parameters can also include Fourier techniques to alter the appearance of the final viewed image as described in "Introduction to Fourier Optics", J. W. Goodman, McGraw-Hill, 1968. For example, it is very common to optically low pass filter an image before it impinges on a CCD sensor. Removing high frequency data prevents aliasing.

Digital or Analog Signal Processing

For pixelated images, it is possible to alter the captured scene parameters to produce desirable characteristics in the visual reproduction, including the preferred viewed tone reproduction. These transformations may be executed in a digital or analog mode depending on whether the image pixels are stored as digital or analog representations. The principals are the same although the mathematical treatment is different. Ref. 5-10 suggest a number of the techniques used in analog and digital processing.

The digital or analog processing can be executed imagewise or partial imagewise (piecewise) to achieve the desired final viewed image. Electronic processing can be done to a single image, several images, or to an entire set of images. The processing can be done separately by color record or as composite color record processing. Single or multi-dimensional look-up-tables of parameters (LUT's) can be used in the processing to facilitate the desired computations. Alternatively, simple or multi-dimensional matrices can be used in the computational processing as can mathematical expressions. Convolutions may also be incorporated during the processing.

Special transformations of the types just described are necessary to achieve the visual reproduction when using a half-tone printing procedure to create the final image.

Image representation compression and decompression may be involved in the digital or analog processing. The image representation may also be transmitted between different transformation agents during the processing procedure.

Digital or analog modes of electronic processing on pixelated recorded images can be effected on single capture scenes, or a series of captured scenes, or on a sequence of related captured scenes to produce the best image or images. It is often possible to create a more pleasing visual reproduction in terms of color balance by analyzing a series of captured scenes as taught by Ref. 20.

The most desirable electronic processing may be determined by an automatic, semi-automatic, user-selected, operator-interactive, or learning algorithm. To create the preferred viewed tone reproduction, a specific tone mapping must be incorporated in the algorithm based on the known tone mapping from the original scene to the input pixelated image representation and the tone mapping inherent in the print generation hardware and media during the final image generation process. However, other desirable electronic enhancements such as color and sharpness boosting may also be included in the electronic manipulation.

The most preferred visual reproductions of the original scenes will depend on the dynamic range of original scene. As taught by the tone mapping criteria in this application, the window of tone mappings provide the preferred representations of the original scene, but the most preferred tone mapping for the visual reproduction will fall within the prescribed boundaries, but its actual path may depend on the original scene. For example, hazy scenes will in general require a tone mapping near the upper boundary of the criteria for the most preferred reproductions, while brightly and clearly illuminated subject matter will in general require a tone mapping near the lower boundary of the criteria for the most preferred reproduction. Selecting the best tone mapping within the specified criteria will therefore require a classification algorithm.

Successful classification algorithms will take many forms, including but not limited to histograms, ranges, parameters based on the distribution, or transformations of the distribution of all or a subset of the recorded or transformed image pixel values. For example, one may define the dynamic range of the original scene by the log exposure density difference between the 5th and 95th percentile log exposure densities for the cumulative pixel log exposure density distribution. If the log exposure density difference is greater than 1.5, the best preferred viewed final image reproduction will have a mapping nearer the bottom of the defined preferred mapping window. If the difference is less than 1.5 the best preferred viewed final image reproduction will have a mapping nearer the top of the defined preferred mapping window.

In digital imaging printing systems, classification algorithms can be implemented to select slightly different tone mappings to create the most preferred images. The input for the classification can be scene parameters or capture conditions. It is thus possible to implement selected tone mappings to create the most preferred images during optical printing by using the typical coarse or fine print balance scanning or by using a preliminary scanning to obtain information to select from among two or more print materials having different densitometric characteristics. This process would require a printer with two or more print material transport systems. It is also possible to employ a single print material with densitometric characteristics that vary depending on some capture condition or controllable parameter such as temperature, exposure duration, light source, zoom status, etc. to produce the most preferable scene reproduction in a single print material printer. Naturally, both methods can be combined in multiple print material printers to produce a greater resolution in achieving the most desirable reproductions.

Information accompanying the captured original scene parameters that describes the camera parameters responsible for capturing the scene can provide useful input for the signal processing algorithms. Useful information includes any single or any combination of scene illumination type, flash parameters such as flash output and/or whether the flash was directed at the subject or bounced onto the subject and/or whether the sufficient flash power was available to properly illuminate the subject, camera lens f-stop, camera exposure time, scene orientation. This information may be used by the camera to optimize the exposure of the film or sensor as taught by Wheeler or Baumeister (Ref. 21 or 22) to produce the best scene capture for use by the signal processing step.

Another part of signal processing is the determination of the proper visual reproduction color balance. This is done for any kind of image to produce the most desirable representation. The final representation color balance may be determined by processing in the camera, scanner, processing computer, or viewed reproduction creation device. For typical photographic film processing, the scanning used to determine print color balance is not usually part of but can be a pixelation process used to produce an electronic image representation. Scene balance algorithms are discussed further in Ref. 23.

Rendering

Rendering is the act of transforming the input electronic, magnetic, or optical image representation to account for the scene parameter manipulation characteristics of the output device and media so that the desired viewed representation is created. Rendering can use chemical, magnetic, optical, electronic, or biological means to accomplish the desired transformation. Rendering can be part of signal processing, part of the final image forming process, or a separate step in the image reproduction forming process.

As an example of rendering, the result of standard PHOTO CD signal processing of a scene derived by scanning a color negative film original capture of original scene parameters, plus additional processing may create a digital electronic representation of the preferred viewed image reproduction described herein. To produce a thermal dye transfer hardcopy representation of the electronic representation, the PHOTO CD representation must be modified in a way which exactly compensates for the non-linear characteristics of the thermal dye transfer process. This includes generating an electrical version of the scene representation which will be applied to the thermal heating elements of thermal printing device to create the dye transfer. Since dye transfer efficiency is not a linear function of applied voltage to the heating elements, the electrical representation must account for this non-linearity to produce the desired preferred viewed scene reproduction. See Ref. 12 for a detailed discussion of rendering.

Step (3)--Creation of the Visual Reproduction Conventional Optical Printing

The visual reproduction of the original scene can be generated by optically printing a photographic film onto silver halide photographic or other light sensitive paper. The printing process can be additive or subtractive and the photographic materials can be negative or positive working (negative film with negative paper or positive film with positive paper). To achieve the preferred final viewed reproduction, the paper response characteristics must be accounted for in the film to be printed so that the resultant produced image has the desired tone reproduction characteristics. For more information on conventional optical printing, please see Hunt (Ref. 1). One suitable embodiment is to capture the original scene parameters with a 35 mm SLR using Kodak Gold+100 film. The film is exposed onto a positive working intermediate film designed to produce the preferred final print reproduction when printed onto Kodak Edge paper using a Kodak Create-a-print processor. The final preferred reproduction is created by printing the intermediate film onto Kodak Edge paper. It is also possible to print onto semi-transparent materials, such as Kodak Duratrans, or on transparent film designed for projection or illuminated display viewing.

Electronic Printing onto Light sensitive Materials

Silver halide photographic or other light sensitive materials are conveniently used by electronic printing processes to produce high-quality reproductions. The light sensitive material can be transparent film, reflective paper, or semi-transparent film. These materials are exposed by visible or infrared light derived from many different sources. The materials may be designed for typical photofinishing applications or they may be specially designed for digital printing applications, such as the paper which is used by specially designed three-infrared light printers manufactured by 3M Company. The photo-sensitive materials respond primarily to three different spectral regions of incident light. Typically, these are red (600-720 nm), green (500-600 nm), and blue (400-500) light. However, any combination of three different spectral sensitivities can be used. These could include green, red, and infrared light or red, infrared 1, and infrared 2 light, or 3 infrared lights of different wavelengths. Or a material sensitive the three primary wavelengths of visible light may be false sensitized so that the color of the exposing light does not produce image dye of the complementary hue, such as red, green, and blue sensitivity producing magenta, yellow, and cyan dye, respectively. Printing can be effected by exposing all pixels sequentially, by exposing a small array of pixels at the same time, or by exposing all the pixels in the image at the same time.

Devices which can be used to print on light sensitive materials include CRT, LED (Light Emitting Diode), LVT (Light Valve Technology), LCD, Laser, as well as any other controlled optical light generating device. See Ref. 2 for a full discussion of these devices. All these devices have the ability to expose 2 or more light sensitive layers in a light sensitive material to produce a colored image. They differ mainly in the technology on which the devices are based.

CRT printers use an illuminated phosphor screen which to expose photographic materials by contact or lens imaging. Typically, 3 colors of exposing light are exposed sequentially by using color filters to gain exposing light purity. The exposing process can be executed in a single step or any combination of steps. Examples of CRT printers include the AGFA Digital Print System and the KODAK PCD 6600 digital print system.

LED printers generally consist of 3 light emitting diodes of different wavelengths which are capable of selectively exposing 3 different color records in a photographic material. Examples of LED printers include early versions of the FUJI Pictography system.

Light valve technology (LVT) can be used to write pixelwise onto Light sensitive materials. These devices are based on 2 or more colored light sources, such as filtered tungsten bulbs, the output of which is focused into a narrow beam and rapidly modulated by means of a light valve. A light valve is an optical material which changes refractive index radically when a voltage is applied. The filtered light selectively exposes different light sensitive layers in the light sensitive material. Examples of LVT printers include the Light Valve Technologies Model 1620B.

LCD devices can also be used to generate color pictorial electronic images which can be used to expose hardcopy photosensitive materials. Liquid crystal displays generally contain a sandwich of polarizers and the liquid crystals whose transmissivity is controlled electrically, although alternative technologies such as Raychem's dispersed liquid crystal materials function in a like manner without polarizers. Other alternate LCD technology applications are disclosed in "Field controlled light scattering from nematic microdroplets", Appl. Phys. Lett., 48, p27, 1986. The colored filters may be arrayed in small adjacent domains, or they may be arrayed depthwise (stacked) one above the other. The LCD controller typically sets/adjusts the red, green, and blue transmissivities associated with a single pixel appropriately to display the input color and brightness. This modulator of light can be used to expose light sensitive materials pixelwise to create the visual reproduction as described in Ref. 28.

Laser printers consist of two or more narrow emission lines derived from one or more lasers which selectively expose color sensitive layers in a light sensitive material to produce a colored image. The lasers may be of any kind. Gas lasers are often used for high speed printers due to their high light output. An example of color laser printers include the FUJI Pictography 3000 printer which uses 2 infrared and 1 red laser diode to expose a specially sensitized paper.

Electronic Printing onto Non-Light sensitive Materials

Non-light sensitive imaging materials are conveniently used by electronic printing processes to produce high-quality reproductions. The printing process can be based on many technologies. The method of image formation can be half-tone, continuous tone, or complete material transfer. The imaging material can be transparent film, reflective paper, or semi-transparent film. The materials can be written on to produce pictorial images by thermal dye transfer, ink jet, wax, electrophotographic, or other pixelwise writing technique. These processes use two or more colorants to create colored pictorial representations of pictorial scenes. The colorants may be dyes, toner, inks, or any other permanent or semi-permanent colored material. See Ref. 2 for a detailed discussion of this kind of printing.

Thermal dye transfer printing is based on the transfer of dyes by pixelwise heating. A thermal printer consists of a controller capable of receiving pixelated image information and controlling the machine in a way which creates a colored image pixelwise as required by the input information, a mechanism to create heat pixelwise, 2 or more dye materials of different hues which can be transferred to the imaging material based on the amount of heat applied, and an imaging material which absorbs or adsorbs the imaging dyes in a controlled manner. The heat may be generated electrically by resistive or other heating elements, by light absorption, or by any other process which generates heat. The printing device may create images by writing pixels individually, by row, or by area. A suitable embodiment for pictorial imaging is the Kodak XL7700 thermal dye transfer printer which uses a resistive head to sequentially transfer a row of pixels at a time from 3 dye-containing thin films (cyan, magenta, and yellow) to a reflective paper receiver. Ref. 2 provides more details.

Ink jet printing uses 2 or more sources of pigments or dyes dispersed in a solvent (inks) to produce a pictorial image on a receiver sheet (reflective paper, transparent film, or semi-transparent film). The pigments are transferred to the imaging material by microdrops pixelwise by using ultrasonic, piezo-electric, or other mechanisms. The final image can be created by sequential applications of individual pixels, by rows of pixels, or by an array of pixels applied together. Suitable embodiments for ink jet printers are the Hewlett Packard HP1200C and the Canon CJ10.

Wax printing is based on the transfer of dyes by pixelwise heating. The transfer is typically done using a half-tone process. A wax printer consists of a controller capable of receiving pixelated image information and controlling the machine in a way which creates the wax pixelwise required by the input information, a mechanism to create heat pixelwise, 2 or more wax pigment materials of different hues which can transferred to the imaging material based on the amount of heat applied, and an imaging material which absorbs or adsorbs the imaging dyes in a controlled manner. The heat is generally produced electrically by resistive elements but can be produced by any other process which generates heat. The printing device may create images by writing pixels individually, by row, or by area. A suitable embodiment is the Tektronix Phaser 200 printer which uses a resistive head to sequentially transfer a row of pixels at a time from 4 wax-containing thin films (cyan, magenta, yellow, and black) to a reflective paper receiver. Ref. 27 provides further derails.

Electrophotographic printing relies on pixelwise electrostatic application of colored toner particles. Two or more hues of toner particles must be applied sequentially to create a color pictorial image. A photoconductor is exposed sequentially for each color record to be written. After each exposure, the photoconductor picks up toner in proportion to the exposure due to the charge attraction between toner particles and the photoconductor. The toner is then transferred to the paper and fused to the paper with heat and pressure. A suitable embodiment for an electrophotographic printer is the Canon CLC500.

Graphic Arts printers for pictorial imaging generally use half-tone printing. These devices use 4 or more inks applied sequentially to produce continuous appearing images. A suitable embodiment for graphic arts applications is the KODAK Approval digital color proofer which uses lasers to locally heat and transfer 4 dyes to a receiver sheet. A halftone process is used in this device. For low volume printing, electrophotographic processes are frequently used. An example is the KODAK Color Edge Copier System. These electrophotographic prints can use 3 or more inks.

Electronic Soft Display

In addition to hardcopy viewed images, it is also possible to create projected images which have the same preference over previously generated pictorial images. Many technologies are appropriate for this kind of image generation. All these techniques rely on producing color images with 2 or more colored lights. These are typically red, green, and blue in nature although they can be any set of primaries as described in Hunt (Ref, 1). Devices which can be used to create the preferred viewed reproduction include CRT, LCD, EL (Electro-Luminescence), LED, light bulbs, lasers, plasma display panels, or any other 2 or more colored lighting apparatus capable of pixelwise illumination. The images can be created by display within the device, projection, or backlighting. Many devices create an image on a screen or display area which is physically a part of the mechanical unit. However, images can also be created by optically projecting the image in the form of light rays from behind or in front of the viewer toward a screen which is in front of a viewer or by projecting a reversed image toward the viewer onto a screen between the viewer and the projecting device.

CRT display uses two or more phosphors which are excited pixelwise by an electron beam. CRT tubes consist of one or more electromagnetically controlled electron beams of small (pixel) size which are displaced spatially and whose intensity is modulated to create a pictorial image by selectively exciting red, green, and blue phosphors coated on a screen at one end of the tube. The incidence of the beam onto the phosphor coating is controlled by a mask so that the red, green, and blue phosphors are generally illuminated by the red, green, and blue electron beam controllers. For more information on CRT devices, see Ref. 13, 28. A suitable embodiment is a Sony Trinitron CRT.

LCD devices can also be used to generate color pictorial electronic images. Liquid crystal displays generally contain a sandwich of polarizers and the liquid crystals whose transmissivity is controlled electrically, although alternative technologies such as Raychem's dispersed liquid crystal materials function in a like manner without polarizers. Systems described in "Field Controlled Light Scattering from Nematic Microdroplets", Appl. Phys. Lett., 48, p27, 1986, may be employed. The colored filters may be arrayed in small adjacent domains, or they may be arrayed depthwise (stacked) one above the other. The LCD controller typically sets/adjusts the red, green, and blue transmissivities associated with a single pixel appropriately to display the input color and brightness. This modulator of light can be used to expose light sensitive materials pixelwise to create the preferred viewed reproduction. The LCD may be as described in Ref. 13, 28, and a suitable embodiment is a Sharp LCD screen.

LED technology can be used to generate pictorial images by projecting 3-color images. LED display devices generally consist of 3 light emitting diodes of different wavelengths, typically red, green, and blue. An example of an LED display is the emissive display by Planar Corp.

Plasma technology can also be used to generate pictorial images. Plasma screens have pixels composed of gas-filled cells. A voltage is applied variably to each cell to produce light. These kinds of devices are often composed of UV-plasma generating cells also containing a visible light phosphor of the desired color. For color images, three different colored phosphors are used in the cells composing the screen. An example of a plasma display is the IBM monitor.

Testing

A test procedure for determining the tone reproduction of an image reproduction system in the case of hardcopy reflection prints will be described in reference to FIG. 13. In this test procedure, two test targets 100, 102 of uniform size are provided. Target 100 is a spectrally uniform gray, i.e. exhibits constant percent reflectance (20%) in a wavelength spectrum of from 380 nm to 780 nm. Target 102 is the same as target 100 except for the provision of a series of uniform gray stepped density areas 102a spanning the range from approximately 0.0 optical density (100% lambertian diffuser) to 1.80 optical density in increment of 0.10 or finer. Each density step is spectrally uniform and is large enough to allow for easy measurement. Both targets are large enough so that when photographed as described below, each target substantially fills the image capture area of the capture device.

A lighting system 103 is provided to uniformly illuminate the targets, mounted on a uniform gray (20% reflectance) background, at approximately a 45.degree. angle of incidence. The lighting should provide reasonably specular illumination typical of high quality, low flare viewing conditions. The spectral quality of the lighting should be similar to that for which the imaging system under test is designed. Under constant illumination conditions from lighting system 103 and with a scene capture apparatus 106, e.g. a photographic camera, oriented perpendicularly to the targets, each of the target images is captured according to ISO standards for the image capture device. Additionally, the absolute visual densities of each density step in target 102 and the corresponding area of target 100 are measured using a very low flare radiometer. Each measurement is made with a spot size one-fourth as large as the density step area being measured. Using identical lighting, image device and radiometer conditions, target 100 is captured and measured as described above.

Using the imaging system 104 under analysis including scene capture apparatus 106 and image reproduction stage 110 and having an overall transformation characteristic represented by transformation box 108, a hardcopy reproduction of the target images is produced by image reproduction stage 110. The reproduction is made in such a manner that a 1.0 scene density relative to a 100% diffuse reflector is reproduced at a density of 1.0.

The reproduced prints are uniformly illuminated with lighting system 103 at a 45.degree. angle of incidence and the visual step densities are measured with the very low flare radiometer 116. It will be understood that the targets and reproductions preferably are illuminated and measured under identical conditions. After correcting the measured step densities on the target 102 for any target illumination non-uniformity using the target 100 densities in the same location as the target 102 steps and after correcting the measured step densities on target 112, the reproduction of target 102, for any target illumination non-uniformity, any field exposure non-uniformity by the scene capture apparatus 106 onto the film or sensor and any field exposure non-uniformities present in the hardcopy image reproduction apparatus 116 using target 114, reproduction of the target 100, densities in the same location as the target 102 steps and using the print through tone reproduction D-LogE function, the measured hardcopy step density versus the target step density function are plotted on a density versus scene density graph along with the instantaneous gamma of this curve. This is preferably done by accurately fitting the data points with a mathematical spline function to obtain the D vs LogE curve and calculating the first derivative of the fitted curve to get the instantaneous gamma function. The resulting plotted functions are then observed and analyzed as shown in FIGS. 5-7.

To test the tone reproduction of video images or projected images viewed in darkened environments, the targets are captured and measured as described above. The images are then reproduced as the intended video or projected images and then measured with a radiometer as described above except using the viewing environment typical for image viewing for the system being tested with no additional lighting. Corrections are then applied for the display hardware field non-uniformity using the photographed (captured) gray target. Plotting and analysis of the resulting D-LogE and instantaneous gamma functions then proceeds as described above except comparison to the parameter of the invention are made using the equivalent adjusted boundary conditions and 100% scene white point as described earlier.

(II) Color Enhancement

The colorfulness of optically-printed photographic prints can be greatly improved by employing high enhancement film with special color papers offering the tone mapping of the invention. Prints produced in this manner successfully exhibit highly saturated color reproduction of highly saturated colors in the original scene, while maintaining preferred skin tone reproductions.

These prints can also be produced digitally using both analog and digital input streams, where the original scene can be digitally printed onto any color paper or where the original scene is captured on film, and the film is scanned and then digitally printed. In every case the tone scale must be modified to match the preferred position and the colorfulness of the scene must be electronically enhanced. From an image capture perspective, the original scenes might exist digitally using either magnetic or optical computer storage systems, on different computer networks, or can be captured and stored independently using electronic cameras (2D, 3D, Video) with CCD sensors. The scenes can exist in analog format and be converted to digital format and stored as an internegative, interpositive, or in an intermediate color space such as RGB, CIELAB or any other defined interchange space. The preferred tone mapping and color enhancement can occur through proper manipulation of scenes captured through reflection scanners, film scanners, or from images encoded and stored on a KODAK PHOTO CD.TM. system.

While preferred tone reproduction typically takes the form of one-dimensional lookup tables, color enhancement generally takes the form of scalars, matrices, polynomials, or multi-dimensional lookup tables. These enhancements are not required to be performed in a predetermined manner. For example, the scalars could be applied before or after a linear, nonlinear, polynomial, or lookup table conversion. Those skilled in the art will recognize that the change in the coefficients of the transforms will reflect the sequence of the applied enhancement.

The combination of preferred tone mapping and color enhancement can be reproduced in any known manner such as by using additive or subtractive hardcopy output devices, or softcopy display. These can take the form of a Cathode Ray Tube, Charged Coupled Device, Light Emitting Diode, LVT, Image Bar, Laser, Dye Sublimation Thermal Transfer, Ink jet, Electrophotographic, or other devices known in the art. Direct thermography where the color change is due to chemicals existing in the paper can also be employed as well. Output systems such as Motion Picture or View Films can also be utilized as well as optical or magnetic Storage Discs. The enhancement may also be displayed using self-illuminating or softcopy output devices such as a Video Display, Liquid Crystal Display, Light Emitting Diode, Organic Electroluminescence Device, projection transparency or backlit projection.

Reproductions obtained in accordance with the invention are such that if the Macbeth Color Checker were included in the original scene, the color of at least one of the six high chroma patches would be enhanced as measured by a CIELAB chroma ratio of greater than 1.0/A. Reference to the "six high chroma color patches" is intended to mean patches 13 through 18 named "blue", "green", "red", "yellow", "magenta", and "cyan" located directly above the grey row as shown in Photographic Materials and Processes L. Stroebel et al, Focal Press 1986, pp 541-545.

To determine whether a transforming process meets the color enhancement metric of the invention, one performs the following steps:

1. Capture the Macbeth Color Checker with the camera normal to the checker and the illumination oriented 45 degrees onto the checker.

2. Generate an Initial Grey Balance Reproduction with the criterion that the CIELAB parameters of the N/3.5 Grey patch in the Initial Grey Balance Reproduction match those of the original N/3.5 Grey Patch.

3. Determine the CIELAB chroma (c*) in the Initial Grey Reproduction for the Light Skin Patch.

4. Generate the Final Reproduction using the criteria that:

(a) the CIELAB chroma (c*) for the light skin patch in the Final Reproduction substantially matches that in the Initial Grey Balance Reproduction;

(b) the CIELAB hue [arctan(b*/a*)] of the Light Skin Patch in the Final Reproduction matches that of the original Light Skin Patch; and

(c) the CIELAB L* of the N/3.5 Grey patch in the Final Reproduction matches that of the original N/3.5 Grey patch.

5. Determine CIELAB chroma values for the 6 high chroma color patches (the row above the gray scale) and for the dark skin and light skin patches. The CIELAB chroma ratios, which are the ratios for a given color patch of c* (Measuring Color R. W. G. Hunt, Wiley (1987), p66) of the reproduction prepared in accordance with the preceeding steps to the c* of the original scene, are calculated for each of these 8 patches. The ratio for at least one of the six high chroma patches should be greater than 1/A and either the CIELAB chroma ratio for the light skin patch is between 0.8/A and 1.4/A or the CIELAB chroma ratio for the dark skin patch is between 1.0/A and 1.8/A or the CIELAB chroma ratios for both of these skin patches is within these ranges. As described with respect to tone mapping, A is 1.0 in the case of direct view reflection media reproduction, 1.1 in the case of self-illuminated reproduction for viewing with ambient surround, and 1.3 in the case of a reproduction for viewing with dark surround.

The most preferred images have ratios for the light skin patch greater 1.00/A and less than 1.4/A and for the dark skin patch greater than 1.0/A and less than 1.8/A.

CIELAB chroma values and grey reproduction exactness are calculated for the CIE standard illuminant D50 from spectrophotometry following CIE recommendations in accordance with CIE Publication No. 15.2 of 1986. Analyzing the reproduction of skin tones and the basic colors, red, green, blue, cyan, magenta, yellow, are indicative of the quality of the most preferred images. In colorimetric terms, at least one of the skin tone patches and the six high chroma patches should adhere to the metric as described. The test metric for color enhancement depends on specific patches of the Macbeth Color Checker. It is appreciated that it is possible to adopt methodology, especially in digital processing, that would have the effect of circumventing the test metric by giving special low- or non-modifying treatment to the particular colors corresponding to these particular patches while modifying the colors in nearby color space in order to accomplish the advantageous result. It is understood that such discontinuous modification methods are intended to be encompassed by the invention.

The Macbeth Color Checker is used as a reference material since it is widely available. We find that the most preferred images are created when the chroma of the skin tone patches is enhanced relative to the chroma of the original within a particular range while at the same time most or all of the six high chroma patches are reproduced with high chroma and at least one of these basic colors is reproduced with a CIELAB chroma ratio greater than 1/A where A depends on the type of reproduction and is as defined for tone mapping.

An extensive set of photographic simulations has shown that when preferred tone reproduction and color enhancement are applied to digitized images, the characteristics preferred by the customer result. In conjunction with the preferred tone reproduction curve, color enhancement can take the form of scalars, matrices, polynomials, or lookup tables. This may take place with regard to the associated color space or output device in a single or multi-step process that can include conversions, gamma correction, half-toning, and/or scanner conversion.

FIG. 14 shows in block diagram form color image reproduction apparatus in accordance with a preferred embodiment of the invention. A CD-ROM input device serves for reading images encoded and stored on a Kodak PHOTO CD disk, and produces image-bearing signals in digital form for each picture element of the image being read. A computer-based workstation which receives the image-bearing signals from the CD-ROM transforms the input image-bearing signals into a format suitable for viewing on a video monitor. Control apparatus which may include a keyboard and cursor enables an operator to provide image manipulation commands pertinent to modifying the video image displayed and the reproduced image to be made. An output device which can be a film writer, thermal, ink jet, electrostatic, or other type of printer, or electronic output device then receives image-bearing signals from the workstation for output onto the appropriate image-receptive medium.

The image data transforms required to convert the image-bearing signals stored on the PHOTO CD disk into image-bearing signals appropriate for creating a reproduced image on the selected output device can be performed in the workstation or computer hardware resident in the output device, etc.

Referring to FIG. 15, one method for converting the image-bearing signals stored on the PHOTO CD into image-bearing signals for creating reproduced images having the preferred tone reproduction and color enhancement characteristics described herein is as follows:

1) The luma and chroma image bearing signals stored on the PHOTO CD disk are converted to nonlinear RGB image bearing signals using the inverse encoding equations found in the publication "KODAK PHOTO CD Products, A Planning Guide for Developers."

2) The nonlinear RGB signals from step 1 are converted to linear RGB signals according to the inverse transformation equations found in the publication "KODAK PHOTO CD Products, A Planning Guide for Developers."

3) The RGB signals from step 2 are converted to CIE XYZ tristimulus values using a 3.times.3 matrix. Since the RGB signals of step 2 are referenced to the CCIR 709 reference primaries, the methods for determining these matrix coefficients are known to those skilled in the art.

4) The XYZ values from step 3 are converted to CIE L* a* b* values using equations published previously by the CIE.

5) The L* values of step 4, which relate to the original scene, are transformed through an appropriate one-dimensional lookup table such that the original scene L* neutral values are transformed to the L* values corresponding to a preferred tonal reproduction.

6) The a* b* values of step 4 are adjusted through scalars, polynomials, or other nonlinear mapping functions so as to enhance the saturated colors while maintaining a preferred skin tone reproduction.

7) The L* a* b* values from steps 5 & 6 are transformed to device driver code values for producing a print on the selected output device. The printer calibration software ensures that the L* a* b* values of the reproduction match the L* a* b* input values.

Referring to FIG. 16, another method for converting the image-bearing signals stored on the PHOTO CD disk into image-bearing signals for creating reproduced images having the preferred tone reproduction and color enhancement characteristics described herein is as follows:

1) The chroma image bearing signals stored on the PHOTO CD disk are adjusted through scalars, polynomials, or other nonlinear mapping functions so as to enhance the chroma values of saturated colors while maintaining preferred chroma values for skin tones.

2) The luma image bearing signal stored on the PHOTO CD disk and chroma values from step 1 are converted to nonlinear RGB image bearing signals using the inverse encoding equations found in the publication "KODAK PHOTO CD Products, A Planning Guide for Developers."

3) The nonlinear RGB signals from step 2 are converted to linear RGB signals according to the inverse transformation equations found in the publication "KODAK PHOTO CD Products, A Planning Guide for Developers."

4) The RGB signals from step 3 are converted to CIE XYZ tristimulus values using a 3.times.3 matrix as described in step 3 of the previous embodiment.

5) The XYZ values from step 4 are converted to CIE L* a* b* values using equations published previously by the CIE.

6) The L* values of step 5, which relate to the original scene, are transformed through an appropriate one-dimensional lookup table such that the original scene L* neutral values are transformed to the L* values corresponding to a preferred tonal reproduction.

B) The L* a* b* values from steps 5& 6 are transformed to device driver code values for producing a print on the selected output device. The printer calibration software ensures that the L* a* b* values of the reproduction match the L* a* b* input values.

Referring to FIG. 17, another method for converting the image-bearing signals applies to a reversal film scanner. The scanner calibration ensures that the scanner output XYZ values match the XYZ values of the film. The image-bearing signals for creating reproduced images having the preferred tone reproduction and color enhancement characteristics described herein is as follows:

1) The scanner output XYZ values are converted to CIE L* a* b* values using equations published previously by the CIE.

2) The L* values from step 1, which relate to the film, are transformed through an appropriate one-dimensional lookup table such that the original film L* neutral values are transformed to the scene L* values for a typical reversal film.

3) The L* values from step 2 are transformed through an appropriate one-dimensional lookup table such that the scene L* values are transformed to the L* values corresponding to a preferred tonal reproduction.

4) The a* b* values of step 1 are adjusted through scalars, polynomials, or other nonlinear mapping functions so as to enhance the saturated colors while maintaining a preferred skin tone reproduction.

5) The L* a* b* values from steps 3 & 4 are transformed to device driver code values for producing a print on the selected output device. The printer calibration software ensures that the L* a* b* values of the reproduction match the L* a* b* input values.

It will be recognized by those skilled in the art that other color conversions are possible for example CIELUV values, or other color spaces used in the art to reproduce the desired effects in a similar fashion. It will also be recognized that the image-bearing signals may be acquired from a print scanner, electronic storage medium, or any other device known in the art.

Referring to FIG. 18, another method for converting the image-bearing signals applies to images stored in scene exposure space. The image-bearing signals for creating reproduced images having the preferred tone reproduction and color enhancement characteristics described herein is as follows:

1) The scene exposure values are logged.

2) The scene log exposure values are transformed to densities using an appropriate one-dimensional lookup table corresponding to the D-Log E curve of the capture film.

3) The density values from step 2 are transformed through a color enhancement matrix.

4) The enhanced densities from step 3 are transformed to Status A paper densities through an appropriate one-dimensional lookup table.

5) The densities from step 4 are converted to CIE XYZ tristimulus values using polynomials, three-dimensional lookup tables or other mathematical means known in the art.

6). The XYZ values from step 5 are converted to CIE L* a* b* values using equations published previously by the CIE.

7). The CIELAB values from step 6 can be isolated, e.g., skin tones, by using three-dimensional lookup tables in order to enhance saturated colors while maintaining a preferred tonal reproduction. Those skilled in the art will recognize that other portions of this or any other color space may be isolated as well.

8). The CIELAB values from step 7 are converted to CIE XYZ values using equations published previously by the CIE.

9). The CIE XYZ values from step 8 are transformed to device driver code values for producing a print on the selected output device. The printer calibration software ensures that the XYZ values of the reproduction match the XYZ input values.

In similar fashion, as will be recognizable to those skilled in the art, steps 1-5 from the previous embodiment can be replaced with a three-dimensional lookup table. While selected areas of color space can be isolated in order to perform color enhancement, an additional subset would be modifying the correlates of color space. An example of this would employ a boost or rotation in the CIELAB hue metric as a form of color enhancement. This would apply to an angular slice of color space as opposed to a non-uniform area isolated by the user, and could be substituted for step 6 of the previous embodiment.

A further example would be the methods disclosed herein where the original image captured on negative or positive photographic films, papers, or electronic capture are transformed to one of the intermediary image data metrics described therein, e.g., relative trichromatic exposures. The encoded images can then be rendered to have the preferred tone reproduction and color enhancement characteristics as follows:

1) The RGB image-bearing signals are converted to logarithmic values and applied to a one-dimensional lookup table which converts the original scene neutral values to those corresponding to a preferred tonal reproduction.

2) The RGB densities derived from step 1 are adjusted through scalars, polynomials, or other nonlinear mapping functions so as to enhance the saturated colors while maintaining a preferred skin tone reproduction.

3) The image-bearing signals from step 2 are transformed to device driver code values for producing a print on the selected output device. The printer calibration software ensures that the image-bearing values of the reproduction match the image-bearing input values.

In accordance with an aspect of the invention, a system is provided for presenting a preferential tone mapping in a reproduction of an original scene, the system being of the type having means for capturing original scene parameters onto a medium from which a reproduction of the original scene is to be derived and means for creating a visual reproduction of the scene from the captured scene parameters. The system is characterized according to the invention by at least one of the capturing and creating means including scene parameter transformation means for imparting to the visual reproduction, when taken in conjunction with untransformed characteristics of the capturing and creating means, a reproduced tone mapping having instantaneous gamma values of viewed image reproduction density relative to density of the original scene is greater than A and less than A plus 0.35 times scene exposure density over a scene exposure density range of from 0.60 to 1.45 as measured relative to a zero density, 100% diffuse reflector in the original scene, the instantaneous gamma values within scene exposure density range further lying entirely within a range of values subtended by an angle of 14.degree. measured from a reference point at 0.0 scene exposure density and a viewed reproduction density gradient of A, wherein A is 1.0 in the case of direct view reflection media reproduction, 1.1 in the case of self-illuminated reproduction for viewing with ambient surround, and 1.3 in the case of a reproduction for viewing with dark surround.

Photographic Film/Optical Printing/Photographic Paper

One of the suitable methods and systems for achieving the preferred viewed hardcopy reproduction uses silver halide color negative film in a conventional or single use camera to capture the original scene parameters (step 1 in the process). The preferred film has densitometric characteristics similar to color negative films used in the marketplace today, having an extended log response region in which density varies linearly with scene log exposure to accommodate over and under film exposure. An additional desirable feature of the preferred film is that it offer strong color enhancement during the chemical development process to produce bright, vivid colors in the final printed image. The transformation step (step 2) is incorporated into the photographic paper sensitometric response so that optical printing of the color negative film onto the color photographic paper (step 3) produces the preferential tone mapping of the preferred viewed reproduction. In this embodiment, steps 2 and 3 are combined into a single event, requiring a specially prepared color paper response function.

Photographic Film/Digitization/Transformation/Digital Printing

Another suitable method and system for achieving the preferred viewed hardcopy reproduction also uses silver halide color negative film in a conventional camera to capture the original scene parameters (step 1 in the process). After film chemical development, the film is scanned to produce a digital representation of the original scene parameters. Suitable scanning is achieved using a KODAK PHOTO CD.TM. scanner.

The digitized image can be stored in this form or the digitized image can be further processed to produce a standard format image, such as that effected by the KODAK PHOTO CD.TM. system, before storing. The KODAK PHOTO CD provides a suitable storage mechanism, although other optical, magnetic, magneto-optical, or other storage modes are equally effective.

Furthermore, the digitized image can be transformed into the preferential tone mapping of the reproduction, either before or after storage. The preferred method for maintaining fidelity to the original scene capture is storage of the original scene capture parameters. For standardization of images from various sources, the Kodak PHOTO CD pathway is suitable, storing the digital images after modification to a standard format. In most cases, it is preferred to apply the transformation (step 2) after the digitized image has been stored since this procedure offers the greatest flexibility to create the desired "preferential tone mapping", which is usually scene dependent.

After original scene parameter capture and image digitization, the image is transformed in step 2 to create the preferred viewed hardcopy image in digital form. This is accomplished based on the densitometric characteristics of the film and the flare characteristics of the camera used in the scene capture. The most preferred transformation will use information contained in the captured scene parameters to determine if the original scene had low, medium, or high dynamic range, and will then produce the most preferred reproduction based on a path through the "preferential tone mapping" region near the top boundary, through the center, or near the bottom boundary, respectively.

While implementing step 3, a rendering correction is applied to the transformed digital image, accommodating the idiosyncrasies of the printer and the printing material, before generating the hardcopy through digital printing. A suitable hardcopy generation step involves color photographic paper, imprinted by a 3-color CRT printer or a 3-color laser printer. These two light-based printing processes offer great speed and very high quality hardcopy images. Any other printing process, such as thermal, wax, ink jet, etc. can also be effective in creating high quality images.

Digital Image Capture/Transformation/Digital Printing

A third suitable method to achieve the claimed preferred viewed hardcopy reproduction uses an electronic digital imaging camera to capture the original scene parameters (step 1 in the process). An example of such a camera is the KODAK DCS electronic camera which captures a scene with a solid state full frame photocapacitor CCD sensor and buffers the image in electronic RAM and stores the image on a magnetic hard disk located in the camera.

The image data in camera RAM may be transmitted and then stored on media, such as magnetic tape, optical disk, magnetic disk, magneto-optical disks, etc. The digital scene parameters can be stored in this form or the image can be further processed to produce a standard format image, such as that effected by the KODAK PHOTO CD system, before storing. The KODAK PHOTO CD disk is suitable as the storage mechanism, although other optical, magnetic, magneto-optical, or other storage modes are equally effective. Furthermore, the digitized image can be transformed into the preferential tone mapping for the reproduction either before or after storage. The preferred method for maintaining fidelity to the original scene capture is storage of the original scene capture parameters. For standardization of images from various sources, the KODAK PHOTO CD pathway is most suitable, storing the digital images after modification to a standard format. In most cases, it is preferred to apply the transformation (step 2) after the digitized image has been stored since this procedure offers the greatest flexibility to create the most preferred tone mapping within the claimed window, which is usually scene dependent.

After original scene parameter capture and image digitization, the image is transformed in step 2 to create the preferred viewed hardcopy image in digital form. This is accomplished based on the response characteristics of the electronic camera and the flare characteristics of the camera used in the scene capture. The preferred transformation will use information contained in the captured scene parameters or capture conditions to classify the "preferred tone mapping" to be used (e.g. whether the original scene's dynamic range was low, medium, or high) and will then produce the most preferred reproduction based on a path through the defined preferential tone mapping region near the top boundary, through the center, or near the bottom boundary, respectively.

While implementing step 3, a rendering correction is typically applied to the transformed digital image, accommodating the idiosyncrasies of the particular printer and the printing material, before generating the hardcopy through digital printing. A suitable hardcopy generation step involves color photographic paper, imprinted by a 3-color CRT printer or a 3-color laser printer. These two light-based printing processes offer great speed and very high quality hardcopy images. Other printing process, such as thermal, wax, ink jet, etc. can be equally effective in creating high quality images.

To assess the preference for reproductions created in accordance with the invention, a series of prints (A=1.0) are made and judged for quality using the following procedure.

A. Light Skin Testing

1. Using a tone reproduction mapping meeting the criteria described herein, a desirable baseline direct view reflection media reproduction position is determined having a light skin tone reproduction falling approximately half way between the Light Skin tone range limits specified for the invention and with at least one of the six basic high chroma Macbeth Color Checker patches (blue, green, red, yellow, magenta, and cyan) reproduced with a CIELAB chroma ratio greater than 1.0.

2. Two reproductions are produced after selecting a scene containing a light-skinned model looking toward the camera, in which at least 50% of the captured scene area is occupied by the model's person above the waist.

a. The first reproduction is prepared using the baseline position described in 1.

b. The second reproduction is prepared exactly like the first except that the light skin tones are reproduced with a CIELAB chroma ratio smaller than the lower limit specified for the invention.

c. A comparison of the two prints reveals a distinct preference by viewers for the print with light skin tones falling within the range of the invention.

d. Similar results are obtained when the remaining five of the six basic high chroma Macbeth Color Checker patches (blue, green, red, yellow, magenta, and cyan) are reproduced with a CIELAB chroma ratio greater than 1.0.

B. Dark Skin Testing

1. Using a tone reproduction mapping meeting the criteria described herein, a desirable baseline direct view reflection media reproduction position is determined having a dark skin tone reproduction falling approximately half way between the dark skin tone range limits specified for the invention and with at least one of the six basic high chroma Macbeth Color Checker patches (blue, green, red, yellow, magenta, and cyan) reproduced with a CIELAB chroma ratio greater than 1.0.

2. Two reproductions are produced after selecting a scene containing a dark-skinned model looking toward the camera, in which at least 50% of the captured scene area is occupied by the model's person above the waist.

a. The first reproduction is prepared using the baseline position described in 1.

b. The second reproduction is prepared exactly like the first except that the dark skin tones are reproduced with a CIELAB chroma ratio smaller than the lower limit specified for the invention.

c. A comparison of the two prints reveals a distinct preference for the print with dark skin tones falling within the range of the invention.

d. Similar results are obtained when the remaining five of the six basic high chroma Macbeth Color Checker patches (blue, green, red, yellow, magenta, and cyan) are reproduced with a CIELAB chroma ratio greater than 1.0.

C. Color Reproduction Testing

1. Using a tone reproduction mapping meeting the criteria described herein a desirable baseline color reproduction position having both light skin and dark skin tone reproductions falling approximately half way between the limits specified in the invention and with the six basic high chroma Macbeth Color Checker patches (blue, green, red, yellow, magenta, and cyan) reproduced with CIELAB chroma ratios between 0.8 and 0.95.

2. Two reproductions are created after selecting a scene containing a blue object with approximately the same CIELAB parameters as the Blue patch in the Macbeth Color checker, in which at least 20% of the captured scene area is occupied by the blue object.

a. The first reproduction is prepared using the baseline position described in 1.

b. The second reproduction is prepared exactly like the first except that only the blue region of color space centered about the Blue patch on the Macbeth Color Checker is reproduced with a CIELAB chroma ratio approximately equal 1.10.

c. A comparison of the two prints reveals a distinct preference for the print with the higher chroma ratio blue colors.

The same results are obtained when the CIELAB chroma ratio for each of the other five basic high chroma colors is increased to 1.1 while maintaining the others in the range of 0.8 to 0.95.

3. Using a tone reproduction mapping meeting the criteria described herein a pleasing baseline color reproduction position having both light skin and dark skin tone reproductions falling approximately half way between the limits specified in the invention, produce eight reproductions after selecting a scene containing a blue, green, red, yellow, magenta, and cyan objects with approximately the same CIELAB parameters as the patches in the Macbeth Color checker having these colors, in which at least 10% of the captured scene area is occupied by each of the six objects.

a. The first reproduction is prepared using the baseline position described in C-1.

b. Reproductions two through seven are prepared exactly like the first except that in each one a different region of color space centered about the particular color (blue, green, red, yellow, magenta, or cyan) patch on the Macbeth Color Checker is reproduced with a CIELAB chroma ratio approximately equal 1.10.

c. The eighth reproduction is prepared exactly like the first except that the regions of color space centered about all six of the basic color patches (blue, green, red, yellow, magenta, and cyan) on the Macbeth Color Checker are reproduced with CIELAB chroma ratios approximately equal 1.10.

d. A comparison of the eight prints reveals a distinct viewer preference for the print with the higher chroma ratios for all of the basic six colors.

Visual reproductions were electronically prepared representing the printing of images captured by color negative film as described in Example 2 and printed onto a paper as described in Example 2 to effect the transformation of the tone mapping and color enhancement image parameters of the invention. Such images were also prepared for paper incorporating only the tone mapping transformation, only the color enhancement transformation, or in a conventional manner employing neither preferred tone mapping or color enhancement.

Nine hundred people were polled as to their quality rankings for a series of such prints of different scenes varying in tone mapping and color enhancement. On average, those polled preferred the reproductions of the invention by 3/4 of a full quality category over the conventional check images. They preferred the reproductions of the invention by a 3/4 category margin over the corresponding reproductions which incorporated tone mapping but not color enhancement. In separate studies, similar polls showed that increasing the color saturation of images prepared on conventional papers (without the preferred tone mapping) resulted in prints judged inferior to those with current levels of color enhancement in the indicated film.

The invention has been described in detail with particular reference to a presently preferred embodiment, but it will be understood that variations and modifications can be effected within the scope of the invention as set forth in the appended claims.